JEWISH PERSECUTIONS AND WEATHER SHOCKS: 1100 1800*

JEWISH PERSECUTIONS AND WEATHER SHOCKS:1100–1800*

Robert Warren Anderson, Noel D. Johnson and Mark Koyama

What factors caused the persecution of minorities in pre-modern Europe? Using panel dataconsisting of 1,366 persecutions of Jews from 936 European cities between 1100 and 1800, we testwhether persecutions were more likely following colder growing seasons. A one standard deviationdecrease in growing season temperature in the previous five-year period increased the probability ofa persecution by between 1 and 1.5 percentage points (relative to a baseline of 2%). This effect wasstrongest in weak states and with poor quality soil. The long-run decline in persecutions was partlyattributable to greater market integration and state capacity.

Throughout most of history, religious minorities were the victims of persecution.Violence against religious and ethnic minorities remains a major problem in manydeveloping countries today (Horowitz, 2001; Chua, 2004; Yanagizawa-Drott, 2015). Incontrast, religious persecution in the developed world is much less common than itused to be.1 This article investigates why some states persecuted minorities more thanothers in pre-industrial Europe and why this persecution gradually decreased between1500 and 1800.

To answer these questions, we focus on the persecution of the Jews in medieval andearly modern Europe. Violence against Jews was caused by many factors, includingreligiously motivated antisemitism. We choose, however, to build on the claimadvanced by historians that Jews were convenient scape-goats for social and economicills.2 To test this hypothesis, we exploit variation in reconstructed city-level temperatureto investigate whether persecutions against Jews were associated with colder temper-atures. We do this using panel data comprising 1,366 city-level persecutions of Jews

* Corresponding author: Mark Koyama, Center for Study of Public Choice, Carow Hall, George MasonUniversity, VA 22030, USA. Email: [email protected].

We are grateful to Megan Teague and Michael Szpindor Watson for research assistance. We benefited fromcomments from Martin Cripps, four anonymous referees as well as Ran Abramitzky, Daron Acemoglu,Quamrul Ashraf, Dean Phillip Bell, Pete Boettke, Tyler Cowen, Carmel Chiswick, Melissa Dell, Dan Bogart,Markus Eberhart, James Fenske, Joe Ferrie, Raph€ael Franck, Oded Galor, Avner Greif, Philip Hoffman, LarryIannaccone, Remi Jedwab, Garett Jones, James Kai-sing Kung, Pete Leeson, Yannay Spitzer, SteliosMichalopoulos, Naomi Lamoreaux, Jason Long, David Mitch, Joel Mokyr, Johanna Mollerstrom, RobinMundill, Steven Nafziger, Jean-Laurent Rosenthal, Jared Rubin, Gail Triner, John Wallis, Eugene White,Larry White and Ekaterina Zhuravskaya. We thank seminar participants at Chapman University, GeorgeMason University, King’s College, London, Stanford University, the Elliott School for International Affairs,the 2013 All UC Conference, the Clio World Congress at the University of Hawaii, Rutgers University,Northwestern University, the Conference on Diversity and Culture sponsored by HSE, Moscow, the DeepDeterminants of Economic Growth Conference at Brown and Colby College. All remaining errors are thefault of the authors. We are grateful to Jane Perry for proof-reading the article.

1 Excepting, of course, the large-scale killings associated with the genocides of the mid-twentieth century.The decline in religious persecution is especially evident when measured on a per capita basis. For a detaileddiscussion of the decline of mass killings and persecutions over time see Pinker (2011).

2 Important historical contributions include Baron (1965a, b, 1967a, b, 1975); Chazan (2006, 2010); Israel(1985); Jordan (1989, 1998); Mundill (1998, 2010); Poliakov (1955); Richardson (1960); Roth (1961); Stow(1981, 1992) and many others.

[ 1 ]

The Economic Journal,Doi: 10.1111/ecoj.12331© 2015Royal Economic Society. Published by John Wiley & Sons, 9600 Garsington Road, Oxford OX4

2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

from 936 European cities that are recorded as having Jewish populations between theyears 1100 and 1800 in the 26-volume Encyclopaedia Judaica (2007) and data on yearlygrowing season temperature (April to September) constructed by Guiot and Corona(2010).

We find that a one standard deviation decrease in average growing seasontemperature (about one-third of a degree celsius) raised the probability that acommunity would be persecuted from a baseline of about 2% to between 3% and 3.5%in the subsequent five-year period. This effect is larger in cities with poor quality soiland with weak states.

To clarify our empirical analysis, we develop a conceptual framework that outlinesthe political equilibrium under which pre-modern rulers would tolerate the presenceof a Jewish community. We show why this equilibrium was vulnerable to shocks toagricultural output and why this vulnerability may have been greater in locations withpoor quality soil and in weak states that were more susceptible to popular unrest.Consistent with our theoretical framework, we find the empirical relationship betweentemperature shocks and persecutions was strongest in regions controlled by weak statesand with poor soil quality.

A growing literature attributes the emergence of modern political and economicinstitutions to the transition from closed to open access institutions between 1500and 1800 (North et al., 2009; Acemoglu and Robinson, 2012). We provide newevidence concerning this transition. In extractive societies, toleration for minoritygroups was always conditional; negative economic shocks led to the unraveling ofthis toleration and to persecutions and expulsions. Our analysis shows that therelationship between negative economic shocks and the persecution of Jews wasweaker in stronger states and diminished as markets became better integrated. Wesupport these results with extensive narrative evidence consistent with thehypothesis that more centralised states as well as better integrated markets playedan important role in reducing the vulnerability of minority groups to economicshocks and in shaping Europe’s gradual transition from closed to open-accessinstitutions.

We follow a number of papers that use weather to identify the impact ofeconomic shocks on violence and conflict in pre-industrial or developingeconomies. Oster (2004) finds that cold weather shocks were associated withwitchcraft trials in early modern Europe. Miguel (2005) finds that extreme levels ofprecipitation were associated with a higher number of witchcraft deaths in modernTanzania. In a historical context, both Zhang et al. (2006) and Tol and Wagner(2010) find a connection between colder weather and warfare in pre-industrialChina and Europe respectively. Bai and Kai-sing Kung (2011) examine therelationship between climate and conflicts between nomads and the state on theborders of pre-modern China. Jia (2014) examines the effect weather-inducedharvest failure had on civil conflict and revolts in Imperial China. Chaney (2013)provides evidence that sharp deviations in the Nile floods strengthened the powerof religious leaders who could threaten the political authorities with revolt. Fenskeand Kala (2013) show that positive climate shocks (a decrease in temperature inthis context) in sub-Saharan Africa increased slave exports and thus had a perverse

© 2015 Royal Economic Society.

2 T H E E CONOM I C J O U RN A L

effect on economic performance and welfare.3 Waldinger (2014) uses temperaturedata similar to those we employ to study the effect of colder temperatures between1500 and 1750 on urban growth in Europe.

By establishing temperature shocks as a potential trigger for persecutions of Jews inpre-industrial Europe, our findings also complement two important recent contribu-tions on the persistence of antisemitic attitudes: Voigtl€ander and Voth (2012) andGrosfeld et al. (2013). Voigtl€ander and Voth (2012) use data from the massacres thatfollowed the Black Death to establish the persistence of antisemitic cultural traits at thelocal level in Germany. They show these cultural factors were an important pre-condition for antisemitic violence in the 1920s and for support for the Nazi party.Grosfeld et al. (2013) examine the persistence of anti-market sentiments in ImperialRussia’s Pale of Settlement where Jews were confined from the end of the eighteenththrough to the early twentieth century. They argue that within the Pale, non-Jewsdeveloped a set of anti-Jewish and anti-market values which have persisted to this day.Our findings are complementary to these studies as we shed light on the timing andcauses of antisemitic violence across Europe between 1100–1800.4 We delineate thecircumstances under which negative supply shocks were more likely to triggerexpropriations and expulsions in societies that were permeated with antisemitism andthus contribute to a growing literature on the causes of persecution.5

Finally, our findings are related to a growing literature on the economic history ofthe Jews in medieval and early modern Europe. Barzel (1992) and Koyama (2010b)study why the Jews were expelled from medieval England. Pascali (2015) examines theconsequences of Jewish expulsions from Italy in the early modern period. He finds thatcities that had either Jewish pawn banks or Monti di Piet�a charitable Christian lendinginstitutions are significantly more financially developed today while those parts of Italythat were ruled by Spain and expelled their Jewish communities in 1541 remaincomparatively underdeveloped. Botticini and Eckstein (2012) provide a novelexplanation for why Jews specialised as moneylenders during the middle ages. Ourtheoretical and empirical results complement their account by explaining why this

3 Dell et al. (2014) surveys the literature on climate and economics. Hsiang and Burke (2014) and Hsianget al. (2013) survey a range of quantitative studies that find a causal link between climate change and socialconflict. Madesam et al. (2013) find that weather has an effect on the ability of political groups in the modernUS to organise. In addition, there are several recent papers which investigate the effect of weather oneconomic outcomes. For example, Jones and Olken (2010) look at the effect of weather on exports, Macciniand Yang (2009) investigate the effect of weather shocks on children’s health, and Dell et al. (2012)investigate the negative effects of warm temperatures on developing economies. Durante (2009) explores therelationship between climatic variability in the past and trust today across European countries. Anderson(2012) finds that lower temperatures were associated with more sentences being passed down by thePortuguese Inquisition.

4 Antisemitism is a nineteenth century term. Nevertheless, following Langmuir (1990), it has also beenused by medieval historians to describe the rise of virulent anti-Jewish hatred and violence after 1100, adevelopment that was based upon a common set of tropes, which sought to blame Jews for personalmisfortunes and tragedies (i.e. in the case of ritual murder accusations) or for general social ills (i.e. in thecharges of host desecration, well poisoning, coin-clipping or diabolism). See Moore (1992, pp. 42–3) andStacey (2000, pp. 163–6).

5 For example, Glaeser (2005) studies the incentives politicians have to incite hatred against particulargroups; Mitra and Ray (2013) provides a theory of ethnic conflict and applies it to Hindu-Muslim violence inIndia; Johnson and Koyama (2014b) examine the relationship between the rise of the French state and thedecline in trials for witchcraft; Vidal-Robert (2013) studies what factors were associated with more trials by theSpanish Inquisition; and Acemoglu et al. (2011) estimate the economic costs of the Holocaust in Russia.


J E W I S H P E R S E C U T I O N S A N D WE A TH E R S HO C K S 3

specialisation, despite making Jews especially valuable to medieval rulers, ultimatelyplaced Jewish communities in a political equilibrium in which they were vulnerable topersecution and expulsion.

1. Weather Shocks and Jewish Persecutions

Jews in early medieval Europe specialised as merchants and moneylenders (Botticiniand Eckstein, 2012, pp. 153–200). Rulers encouraged Jewish settlement in order topromote economic development. The Bishop of Speyer in 1084 wrote: ‘When I wishedto make a city out of the village of Speyer, I Rudiger, surnamed Huozmann, bishop ofSpeyer, thought that the glory of our town would be augmented a thousandfold if Iwere to bring Jews’ (quoted in Chazan, 2010, p. 101).

A political equilibrium emerged that enabled European rulers to benefit financiallyfrom the presence of Jewish communities but which left the Jews themselves vulnerableto persecution and the threat of expulsion. Over time, Jews specialised as moneylen-ders (Botticini and Eckstein, 2012, pp. 201–47).6 Jewish usury was frequentlycondemned by the Church but it was promoted by secular rulers both because creditwas understood to be crucial to the medieval economy and because it provided rulerswith an accessible tax base. The Diet of Mainz proclaimed that, ‘as loans are necessaryand Christians prohibited to lend on profit, the Jew must be allowed to fill the gap’(Stein, 1956, p. 144).7

As outsiders in a society that defined itself in opposition to unbelievers, Jews arousedsuspicion from others.8 But it was in the long-run interest of secular rulers to protecttheir Jewish communities. According to medieval political theory: Jews were serfs ofthe exchequer because, in return for protection against violence, they had submittedto the king and could therefore be taxed at his discretion (Baron, 1967b). Rulersexploited Jews as ‘fiscal sponges’ to use a contemporary metaphor: ‘No sooner didthey suck up the money [from the population through their usury], than theoverlords proceeded to squeeze it out of them into their own pockets’ (Baron, 1967b,p. 199).9 The major problem with this arrangement was that medieval rulers struggledto make credible commitments. They were often unable to protect Jewish communities

6 See Baron (1967b, p. 135). Emery (1959); Lipman (1967); Mundill (1991) and Botticini (1997) provideexcellent empirical studies of how medieval Jewish moneylenders operated. For analysis of the tightening ofthe usury prohibition see Chazan (1973–74) and Koyama (2010a).

7 As Baron (1967b, p. 198) wrote: ‘Many Jews and Christians alike realised that, next to the religioustradition, the main reason for the former’s toleration in western lands was the rulers’ self-interest in therevenue derived from them’.

8 See Moore (2008, pp. 26–42). This hostility could manifest itself among elite groups as well as amongpeasants. In Renaissance Italy, Jewish moneylenders lent to the poor and were often championed by them,and were typically, instead, opposed by city elites (Botticini, 2000). It is important to stress that we do notattempt to provide an economic or rational choice explanation for the virulent antisemitism that emerged inmedieval Europe and that had a variety of sources (Trachtenberg, 1943; Voigtl€ander and Voth, 2012). Forexample, Menache (1985, 1997) analyses the importance of the blood libel myth in generating anatmosphere conducive to expulsion. What we do attempt to explain is why negative economic shocks led tothe expulsion and expropriation of Jewish communities in some polities but not in others.

9 In France, this implicit agreement first appears to have been stated in 1198 when the Jews were readmittedinto the Royal Domain by Philip Augustus (Moore, 2008, p. 41). In England, where it was perhaps most fullydeveloped, it followed themassacres of Jews in York and the establishment of theExchequer of the Jewry in 1194(Cramer, 1940; Stacey, 1985 1995; Dobson, 2003; Brown and McCartney, 2005; Koyama, 2010b).



from unrest or antisemitic violence and frequently faced the temptation to expropriateJewish communities for short-term gain.

The Pastoureaux or Shepherds’ Crusade of 1320 illustrates how temperature shocksdisrupted the political equilibrium that protected Jews in the medieval states. The Jewswere allowed to resettle in France in 1315 on the condition that they act asmoneylenders and fiscal agents for the crown. Unfortunately, a succession of disastrousharvests ensued between 1315 and 1321. Campbell describes this as possibly ‘the singleworst subsistence crisis, in terms of relative mortality, in recorded European history’(Campbell, 2010, p. 7).10 These shocks produced civil unrest across swathes ofnorthern Europe. In France, a movement arose known as the Pastoureaux. ThePastoureaux first attacked royal castles in Normandy and the Paris region, then theymoved south where they persecuted Jews throughout Languedoc in cities like Saintes,Verdun, Grenade, Castelsarrasin, Toulouse, Cahors, L�ezat, Albi, Auch, Rabastens andGaillac (Barber, 1981a, p. 12). The ‘brunt of peasant violence fell upon the Jews, forthey . . . could be blamed for the economic hardships which the lower classes hadrecently been suffering’ (Barber, 1981b, p. 163).11

The king protected the Jews where possible but massacres occurred wherever royalauthority was weak. This example illustrates the connection between a subsistencecrisis, political unrest and antisemitic violence in a state that was unable to protect itsJewish minority. There are many other examples of persecutions that follow thispattern of weak polities being unable to credibly commit to their agreements withJewish communities in the face of economic shocks. In the next Section, we develop aframework that clarifies the mechanisms that connected temperature shocks tosubsistence crises and political unrest and then to antisemitic violence.

2. A Conceptual Framework

In this Section, we develop a simple framework to explore the relationship betweencolder temperature, income shocks, state capacity, and persecutions. Standard modelsof autocratic states struggle to explain why rulers would ever expel or destroy a valuablefiscal resource like the Jews. To understand why expulsions occurred, therefore, weneed to consider cases where the sovereign cannot credibly commit to protect avulnerable minority.12

10 It was followed by the so-called Great Bovine Pestilence, which wiped out 60% of livestock on thecontinent (Slavin, 2012).

11 Tension had been building for sometime as a result of the poor harvests and the peasants undertookreligious demonstrations and parades aimed at ending the famine (Barber, 1981b, pp. 162–3). Contempo-raries also mention that they were incited by debtors of the Jews (Barber, 1981b, p. 146). Nirenberg arguesthat ‘the shepherds and the townspeople who supported them’ understood this relationship, and ‘recognisedthat the heavy taxes placed on Jews were a form of indirect taxation on Christians’ (Nirenberg, 1996, p. 48).When the Pastoureaux attacked Jews and looted their possessions in face of the royal attempts to protect them‘they were both attacking a much-resented aspect of administrative kingship and dramatising the state’sinability to protect its agents, the Jews’ (Nirenberg, 1996, p. 50). Note that this is an instance of peasantsattacking Jews; in other cases, towns people targeted the Jews. This means that it is difficult to generaliseabout which economic strata would have the strongest incentive to expropriate the Jewish community in theevent of a negative shock to the agrarian economy.

12 We build on the reasoning developed by Acemoglu (2003) in our assumption that the Jewishcommunity cannot make a Coasian bargain with the ruler in order to avoid persecution.



Consider an economy in which agriculture is the predominant source of income.Agricultural output is a function of temperature as well as social quality, labour inputsand a host of other variables. In Europe colder growing season temperatures reduceagricultural output. In particular, several successive years of below optimal temperaturecould cause significant hardship and pressure on resources, especially if a dearthcaused peasants to consume their seed-corn. We will refer to such episodes astemperature shocks.13 In sub-Saharan Africa high temperatures adversely affectagricultural output but, in Europe, wheat and other grains can survive cold wintertemperatures but are strongly affected by cold and wet weather during the growingseason.14

We assume that the Jews play an important role in the commercial sector of theeconomy and that they are a valuable source of revenue to the ruler. In normal times,therefore, the ruler imposes a tax on the Jewish community that maximises discountedfuture tax revenue subject to the costs of collection and enforcement. In exchange forthese taxes, he promises to offer protection to the Jewish community from either eliteor popular antisemitism.

Consider a simple political economy framework in which temperature shocks loweragricultural output and incomes and hence cause political unrest. This unrest can bedriven by either peasants or elites. Low incomes can cause peasants to rebel directly.But low agricultural output could also cause a crisis of ‘surplus extraction’ among elites(as low agricultural output reduces the amount of income they can extract via rents,taxes or feudal dues). In these situations, Jews were often targets – either because theywere held to be directly responsible for the misfortune of the population or becausethey were vulnerable targets and perceived to possess large amounts of wealth. As aresult, this framework predicts that periods of colder temperature will be associatedwith a higher probability of persecution.

Rulers of states with high state capacity can quell such rebellions relatively easily. Incontrast, rulers of states with low capacity are highly vulnerable to either popular orelite unrest. Stronger rulers will therefore be able to credibly commit to protectingtheir Jewish community regardless of the income shocks that they face. Weaker rulers,however, will not be able to make such a commitment. Moreover, rulers of states withless developed fiscal capacity will be more likely to face a fiscal crisis as a result ofperiods of sustained low harvests, whereas rulers of states with greater fiscal capacitywill have more access to alternative sources of revenue and revenue-smoothingtechnologies. For states with low fiscal capacity, on the other hand, the easilyappropriable wealth of the Jewish community will be a more tempting target inperiods of fiscal distress. As a result, weaker rulers are more likely to expropriateJewish communities themselves, in anticipation of antisemitic violence or unrest.

13 In general, there is an optimal temperature, sayT �, that solves:Q = F(T; X) whereQ is agricultural outputand X a vector of other inputs. Deviations from the optimal temperature |T � T*| lower agricultural output. Atemperature shock is a deviation that is large relative to the standard deviation of temperature fluctuations.

14 Numerous studies argue that there is a strong relationship between cold temperature and reducedagricultural output in western Europe (see Galloway, 1986, for a survey of this literature). According to Porterand Gawith (1998) wheat has a lethal low temperature of �17:2� 1:2�C , and a lethal high temperature of47:5�C . They note that the optimal temperature for wheat cultivation over the course of the growing season isbetween 17 and 23�C (Porter and Gawith, 1998, p. 25).



A second prediction from this analysis is that the relationship between coldertemperatures and Jewish persecution will be stronger in areas governed by weakstates.

Finally, our framework suggests several other factors that could mediate therelationship between temperature shocks and persecutions. There is evidence thatregions with poorer quality soil might be more vulnerable to periods of coldertemperature as better soil quality mitigates the negative effect of a temperatureshock.15 However, areas with poorer quality soil might also have higher demandfor the lending services provided by Jewish communities. Therefore, rulers mighthave a stronger incentive to protect Jewish communities in these regions. A prioriit is unclear which effect would dominate. Second, market integration and accessto urban markets could dampen the effect of local temperature shocks onagricultural outcome and food price. It should be noted, however, that there werelimits to the ability of markets to smooth shocks to output in the pre-industrialworld as it was expensive to transport foodstuffs.16 Moreover, spatial correlation intemperature shocks limited the ability of markets and trade to smooth volatility inthe price of food staples. Nevertheless, in areas closer to well-developed markets,we expect the relationship between cold temperatures and persecutions to beweaker.

3. Empirical Analysis: The Effect of Temperature Shocks on JewishPersecutions

Our main empirical specification is as follows:

yit ¼ bTi;t�1 þ gi þ lt þ XitXþ eit ; (1)

where yit is a binary variable measuring whether a persecution or expulsion occurred incity i during period t. When we include all violent acts against Jews in the dependentvariable we call this measure Persecutions. When we only include expulsions, thisvariable is labelled Expulsions: Ti;t�1 is a measure of the average temperature for city i inthe period preceding the persecution expressed as the degrees celsius deviation fromthe 1961–90 average. We report both estimates in which we include a full vector of cityfixed effects, gi , as well as the DID estimates, which include time dummies, lt . In ourbaseline regressions, we use five-year averaged data so that each period t refers to anaverage of temperatures over the years in that period e.g. 1535–9; consequently wehave 140 observations for each city.17 We do this primarily because of the potential

15 There are several reasons why we expect regions with poor quality soil to be more vulnerable to climaticshocks. Higher quality soil is more robust to extreme variations in temperature (Porter and Semenov, 2005).Consistent with this, Malik and Temple (2009) find that regions with poor soil quality appear to experiencegreater volatility in agricultural output. In this case, a population in an area with poor soil quality might bemore vulnerable to colder temperatures than a population in an area with high quality soil, even if theirlevels of per capita income were similar.

16 Masschaele provides estimates of transport costs for fourteenth century England. On average,transporting wheat by land would increase the price by 0.4% per mile – doubling the price of grain forevery 250 miles travelled. A ten mile journey raised the price by 4% (Masschaele, 1993, p. 274). Otherevidence suggests that transporting grain was much more expensive in most of continental Europe.

17 That is, seven centuries of data times twenty five-year periods in each century.



measurement error in both the persecutions data and the temperature variable (Guiotand Corona, 2010).18 As such, if the persecution is recorded as occurring in 1541, weuse the data on temperature from 1535 to 1539 to explain it in the five-year data. Wefocus on temperature during the preceding five-year period rather than temperatureduring the contemporaneous period in order to reduce measurement error andbecause historians argue that successive bad harvests placed the greatest strain onagricultural societies.19 In Section 4, we show our results are robust to using yearly datawith city fixed effects and year dummies.20 All regressions include controls for the tenyears surrounding the Black Death in Europe (1346–55) and a measure of urbandensity around city i at time t.21 We use a linear probability model as our mainspecification. We prefer the linear probability model as nonlinear models with fixedeffects suffer from the incidental parameter problem and this can bias asymptoticstandard errors downwards (Greene, 2004).

3.1. Data

In order to measure violence against Jews, we use city–level data on the presence of aJewish community in Europe between 1100 and 1800 taken from the EncyclopediaJudaica (2007). There are 1,069 cities in our complete data set. This number falls to936 after we introduce our urban density control variable.22 Figure 1 illustrates thegeographic coverage of our Jewish city data and the distribution of persecutions for theentire period. The Encyclopedia typically mentions when Jews entered a city, when theywere expelled, when there was some other violent act perpetrated against them, andwhen they were allowed re-entry (if ever). We are interested in all of these pieces ofinformation since in order to model the probability of violence against Jews, we need toknow when that city had a Jewish population.23

Using these data, we construct a variable called ‘Jewish Presence,’ which is equalto one during all the years there is a known Jewish community in the city and zerootherwise. There is a Jewish community present in the average city about 40% ofthe time. Our dependent variable Persecution is binary and measures whether thereis either an expulsion or any other major violent act (i.e. a pogrom) against Jews ina city in a given five-year period. There are 1,366 such events in our base data set:821 expulsions and 545 pogroms. Our other dependent variable is Expulsion and

18 For instance, several medieval edicts of expulsion allowed the Jews a period of time of up to six monthsto leave the city. Thus, for several cities, different sources record different dates of expulsion for a singleevent e.g. the city of Cologne decided not to renew the permit it granted Jews to reside in the cities in 1423,but only enforced the expulsions of the Jewish community itself in 1424.

19 These claims are robustly supported by our placebo regressions which we report in Section 4. There wefind effects of negative temperature on persecution probability up to eight years preceding the event.Reassuringly, we find no evidence that future temperature affects persecution probability.

20 In the yearly regressions, e.g. the persecution in 1541 is explained using temperature data in the five-year period prior to the persecution i.e. 1537 to 1541.

21 See online Appendix D for a discussion of the Black Death and violence against Jews. See onlineAppendix C for further description of the population density measure.

22 We lose some cities in eastern Europe. Excluding the urban density control and including these citiesdid not change results.

23 A more detailed description of all our data is contained in online Appendix C.



measures whether there is an expulsion or not in the city during the five-yearperiod.

We restrict our analysis to use only the sample of cities that currently have a Jewishpopulation. Cities without a Jewish population are treated as missing variables. Thisapproach is consistent with a conventional strategy used in discrete-time survivalanalysis as discussed by Box-Steffensmeier and Jones (2004) and Yamaguchi (1991).Furthermore, our results will also be easier to interpret than with the alternative Coxhazard models.

In order to measure temperature shocks, we employ data on reconstructedtemperature for medieval and early modern Europe provided by Guiot and Corona(2010). Guiot and Corona (2010) collect information from numerous proxy sourcesincluding 95 tree ring series, 16 indexed climatic series based on historical documents,ice-core isotopic series, and pollen-based series to construct a 32 point grid ofreconstructed temperature during the growing season (April to September) for all ofEurope between 900 and the present day. Their historical temperature reconstructionsare based on model mapping proxies into growing season temperatures. This model iscalibrated using actual temperature data from 1850 to 2007. We use geospatial softwareto interpolate the temperature for the area between the grid points so that we have a

Total Persecutions01–89–1516–2627–42

Fig. 1. Jewish Persecutions, 1100–800Notes. Circles represent a Jewish city that has at least one persecution. Larger circles representmore persecutions. Triangles are Jewish cities in our data that never persecuted.Source. Encyclopedia Judaica (2007).



smooth map for each year. Finally, we extract the yearly temperatures for each of ourcities. We follow Guiot and Corona (2010) in expressing the temperature data in termsof differences relative to the 1961–90 average.

To address potential bias stemming from unobserved economic development of acity or its access to markets, we construct a measure of urban density. This approach isconsistent with other work on early modern Europe that relies on urbanisation data inlieu of estimates of per capita GDP or market development.24 Our urban densityvariable is called PopDensity and is based on the Bosker et al. (2013) dataset of cities. Asour Jewish cities do not correspond perfectly to the Bosker et al. cities, we usegeospatial software to create a population heat map for every century based on allBosker et al. cities with populations greater than 5,000. Each point on the map isassigned a population number based on the inverse distance-weighted value of allBosker et al. cities within 1 degree of the point (about 100 kilometres depending onthe latitude).25

One other potential source of bias in our estimates is the disproportionate effect ofthe Black Death on Jewish persecutions in the years immediately following its arrival inEurope in 1348. Figure 2 shows how average temperature and the total number ofpersecutions varied over time. The Black Death is an obvious outlier.26 Jewishcommunities faced unprecedented levels of persecution during the first wave of theplague – as they were singled out as scapegoats and easy targets of expropriation andpopular violence as the existing social order weakened (Voigtl€ander and Voth, 2012).Since we do not want to identify the effect of temperature on antisemitic violence usinga disproportionate number of observations associated with epidemiological causes, wechoose to control for the Black Death years by allowing both the intercept and theslope in all specifications to vary for the years 1346–55.

3.2. Baseline Results

In panel (a) of Table 1, we report the results of estimating (1) using Persecution as thedependent variable in columns (1)–(3) and Expulsion as the dependent variable incolumns (4)–(6). Robust standard errors clustered on the nearest temperature gridpoint from Guiot and Corona (2010) are reported in parentheses.27 We multiply allcoefficients by 100 for ease of interpretation.

In specification (1), we report the b under OLS regression with controls includedbut with no fixed effects. The negative sign implies that colder temperature raises theprobability of a persecution of Jewish communities. The size of the coefficient implies aone degree decrease in five-year average temperature leads to a 1.91 percentage point

24 See De Long and Shleifer (1993); Acemoglu et al. (2005) and Nunn and Qian (2011) among manyothers.

25 We provide more details about the construction of this and other variables in online Appendix C anddescriptive statistics are in online Appendix A.

26 Another potentially influential set of data points are associated with the Iberian national expulsions atthe end of the fifteenth century. We will exclude these from our regressions in panel (b) of Table 1 todemonstrate the robustness of our baseline results.

27 These grid points are reproduced as the red dots in online Appendix C, Figure C6. Cities closer to agiven grid point receive more correlated temperature shocks. In contrast, temperature shocks are a randomwalk over time, thus the temperature-grid seems the appropriate level of clustering.



increase in the probability of a major act of violence against Jews over the subsequentfive years. When we include city fixed effects in specification (2), this probabilityincreases to 2.52 percentage points. Including five-year time dummies in specification(3) increases the DID estimate to 3.19 percentage points.

Although a one degree decrease in temperature is convenient to consider, it is alsoquite large. Under the DID specification in column (3), a one standard deviationdecrease, or 1/3 of a degree, in temperature increases five-year persecution probabilityby 1 percentage points which is equal to 50% of the sample mean of the dependentvariable (2.22). Restricting our attention to explaining just expulsions in columns (4)–(6) of panel (a) of Table 1 yields similar results.

Panel (b) of Table 1 replicates the results in panel (a), after excluding the Iberiannational expulsions. In general, the sizes of the coefficients we obtain are smaller butthey retain statistical significance. Our preferred DID estimator suggests that a onedegree decrease in the temperature of the previous five years is associated with a 1.6percentage points increase in the probability of persecution. A one standard deviationdecrease in temperature increases the probability of persecution by a half a percentagepoint (0.3491.6). Overall, our main results are robust to the exclusion of the Iberianexpulsions.

The results in Table 1 provide strong empirical support for the first prediction of ourmodel: colder temperature disrupted the political equilibria of pre-industrialEuropean societies and increased the likelihood that minority communities wouldbe persecuted. However, on their own, these regressions do not shed light on the

Black Death Iberian Expulsions

–1.0

–0.5

0

0.5

1.0

Tem

p D

evia

tion

(deg

Cel

sius

)

0

50

100

150Pe

rsec

utio

ns

1100 1200 1300 1400 1500 1600 1700 1800Year

Persecutions Temp Deviation

Fig. 2. Jewish Persecutions and Temperature DeviationsSource. Encyclopedia Judaica (2007) and Guiot and Corona (2010).



economic, institutional and political mechanisms connecting negative temperatureshocks to Jewish persecutions.

3.3. Mechanisms

Our conceptual framework suggests that the effect of temperature shocks on Jewishpersecutions should be greatest in societies with weak political institutions. It alsoprovides some reason for expecting differential results in areas with poorer quality soil.Areas with worse quality soil might be more vulnerable to temperature shocks. But thevalue of Jewish moneylending to a ruler might also be more valuable in such regions.In this subsection, we explore these predictions by estimating a series of specificationsbased on:

yit ¼ aTi;t�1 þ bMit þ fMit � Ti;t�1 þ gi þ lt þ X 0itXþ eit ; (2)

where yit is our measure of persecution or expulsion in city i in period t. Mi;t is one ofour mechanisms, such as a measure of soil suitability or state capacity for city i in period

Table 1

Baseline Results

Persecutions Expulsions

(1) (2) (3) (4) (5) (6)

Panel (a): Baseline effects of temperature on persecutions and expulsionsTemperaturei;t�1 �1.91*** �2.52*** �3.19*** �1.59*** �2.06*** �2.31***

(0.625) (0.740) (1.04) (0.447) (0.512) (0.891)Controls Yes Yes Yes Yes Yes YesCity dummies No Yes Yes No Yes YoTime dummies No No Yes No No Yes

N 55,698 55,698 55,698 55,698 55,698 55,698F 5.202 5.609 6.695 4.741 5.705 5.108p-values 0.0029 0.0019 0.0006 0.0048 0.0017 0.0012

Panel (b): Excluding the Iberian national expulsionTemperaturei;t�1 �1.08*** �1.57*** �1.60*** �0.751*** �1.10*** �0.667*

(0.259) (0.382) (0.527) (0.132) (0.232) (0.380)No IberianNational

Yes Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes YesCity dummies No Yes Yes No Yes YesTime dummies No No Yes No No Yes

N 55,253 55,253 55,253 55,253 55,253 55,253F 6.759 6.604 6.214 9.128 6.837 3.566p-values 0.0006 0.0007 0.0012 0.0001 0.0006 0.0284

Notes. * p < 0.10, ** p < 0.05, *** p < 0.01. Standard errors clustered at the climate gridlevel n parentheses.Co-efficients are multiplied by 100 to represent percentage points. Observations are at the city 9 five-yearlevel between 1100 and 1799. In both panels the dependent variable is either Persecution (0 or 1 if either anexpulsion or other violent acts against Jews occurs) or Expulsion. Panel (a) uses the full sample. In panel (b)the Iberian national expulsions in 1492 (Spain) and 1497 (Portugal) are excluded. All regressions include ascontrols a dummy and slope variable for the 10 years surrounding the Black Death and a measure ofpopulation density (see online Appendix C). Co-efficients are reported with standard errors clustered at thetemperature-grid level.



t. For consistency and ease of interpretation, we code the mechanism variable to beeither a 0 or 1 such that 1 corresponds to our theoretically predicted higher probabilityof a persecution (e.g. poor soil or low state capacity).28Mi;t � Ti;t�1 is an interactionvariable of the mechanism with the lagged temperature shock. gi and lt are city andtime dummies respectively, X it is a vector of control variables, and eit is an i.i.d. errorterm. We are interested in three quantities: the effect of temperature on persecutionprobability in cities where the mechanism is not present, a; the indirect effect of themechanism on persecution probability, f; and the total effect of temperature onpersecution probability in cities where the mechanism binds, which is given as,@yi;t=@Ti;t�1 ¼ aþ f�Mi;t , with Mi;t evaluated at 1.

The first mechanism we investigate in Table 2 is whether cities surrounded by lowquality soil were more likely to persecute their Jewish communities. In columns (1) and(2), we test this hypothesis using our preferred measure of agricultural productivity –wheat suitability taken from the FAO (Fischer et al., 2002). This database is constructedby combining characteristics of wheat (optimal growing temperature, soil type etc.)with highly disaggregated climatic and geographic data covering variables such asprecipitation, cloud cover, ground-frost frequency, soil types and slope characteristics.The data have a spatial resolution of 0.5 degree 9 0.5 degree (or about 60 9 60kilometres at 45 degrees latitude, typical for France).29 We extract the wheat suitabilityfor each of our cities using geospatial software and then follow a similar strategy asNunn and Qian (2011) in creating a dummy variable equal to one if a city has anagricultural sector that is either moderately or significantly constrained in its wheatcultivation. This is the main variable Low Wheat that we use in our regressions.

Regressions (1) and (2) show that cities with soil unsuited for agriculture were morelikely to persecute Jews. In the fixed effects regression in column (1), the coefficient onf is 2.77 and statistically significant.30 This accounts for more than half of the overalleffect of temperature on persecutions reported at the bottom of the column as 3.98.Under the DID specification in column (2), the coefficients on the indirect effect oftemperature through soil suitability is not statistically significant. Nonetheless, theoverall effect is precisely estimated and economically significant with a value of 3.66.31

We next test whether persecutions were more likely in cities associated withcomparatively weak polities. To do this, we employ the disaggregated data from theState Antiquity Index of Bockstette et al. (2002). These data include a measure formodern-day countries of whether that area was a stable state for 50-year intervals from1 B.C. to the present day. Specifically, every modern-day country, in each 50-year period,is assigned three numbers. The first is a one if there was a government above tribal leveland a zero otherwise. The second number is a one if the government is locally based,0.50 if it is foreign based, and 0.75 if in between. The third number measures how

28 In online Appendix B Table B1, we provide results using continuous measures of both soil quality andstate antiquity.

29 We assume ‘intermediate’ inputs. Online Appendix C describes these data in more detail.30 The direct effect of Low Wheat is not reported since it is a time-invariant variable and thus absorbed by

the fixed effects.31 We also run our mechanism regressions for soil quality using continuous versions of these variables in

online Appendix B. The results for soil quality clearly show a statistically significant effect for low quality soiland zero effect for high quality soil.



Tab

le2

Mechanisms

Dep

enden

tvariab

le:persecu

tions

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Tem

peraturei;t�

1�1

.21*

*�2

.53*

**�1

.38*

**�2

.20*

**�2

.75*

*�3

.59*

**�3

.03*

*�3

.30*

*(0.485

)(0.876

)(0.429

)(0.738

)(1.01)

(1.24)

(1.36)

(1.64)

Low

wheat

9Tem

peraturei;t�

1�2

.77*

*�1

.13

(1.29)

(0.731

)Low

StateAntiquity

0.76

02.09

***

(0.454

)(0.524

)Low

StateAntiquity9

Tem

peraturei;t�

1�2

.92*

*�2

.34*

**(1.18)

(0.78)

Low

Constraints

�1.43*

*0.32

1(0.611

)(0.699

)Low

Constraints9

Tem

peraturei;t�

11.29

1.36

*(1.02)

(0.758

)Low

Capital

Protection

�1.63*

0.25

9(0.89)

(0.626

)Low

Capital

Protection9

Tem

peraturei;t�

10.87

60.11

3(1.26)

(1.04)

Tem

peraturei;t�

1+Interaction

�3.98*

**�3

.66*

**�4

.30*

**�4

.54*

**�1

.46*

**�2

.23*

**�2

.16*

**�3

.19*

**(1.2)

(1.2)

(1.20)

(1.26)

(0.37)

(0.69)

(0.6)

(0.83)

Controls

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Citydummies

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Tim

edummies

No

Yes

No

Yes

No

Yes

No

Yes

N55

,698

55,698

55,698

55,698

55,698

55,698

55,698

55,698

F6.85

715

.65

5.89

16.25

411

.36

4.27

54.35

15.66

4p-values

0.00

030.00

000.00

000.00

050.00

000.00

520.00

320.00

10

Notes.*p

<0.10

,**p

<0.05

,***

p<0.01

.Standarderrors

clustered

attheclim

ategrid

leveln

paren

theses.O

bservationsareat

thecity9

five-yearlevelb

etwee

n11

00an

d17

99.Thedep

enden

tvariab

leisPersecution

(0or1ifeither

anex

pulsionorother

violentactsagainstJewsoccurs).AllMechan

ism

variab

lesare0or1.

Seetext

andonlineAppen

dix

CfordescriptionsoftheMechan

ism

variab

les.Allregressionsincludeas

controlsadummyan

dslopevariab

leforthe10

yearssurroundingthe

Black

Death

andameasure

ofpopulationden

sity.Tem

perature+Interactionisthetotaleffect

oftemperature

onpersecu

tionprobab

ilitymeasuredas

thesum

ofthe

coefficien

tofTem

perature

plusthat

oftherelevantMechan

ism

interactionterm

.Coefficien

tsarereported

withstan

darderrors

clustered

atthetemperature-grid

level.



much of the territory of the modern government was ruled by the polity during the50-year period.32 These scores are then multiplied by each other and then by 50. Thisresults in a panel of scores for present day countries measuring in every 50-year periodwhether it was an autonomous nation (a score of 50) if it had a tribal level ofgovernment (score of 0) or something in between. Bockstette et al. (2002) thenaggregate these data to get a single score for state antiquity of modern countries. Weare interested in the disaggregated historical data, however. Thus, after interpolatingbetween 50-year periods, we extract their values for each of our cities using geospatialsoftware. We then create a dummy variable for each city equal to one if that region hasa score less than the average for the rest of the sample and zero otherwise. We call thisdummy Low State Antiquity.33

The regression results in columns (3) and (4) of Table 2 support our theoreticalprediction that cities in regions with weaker states will be more likely to persecute Jewishpopulations. The coefficient on the indirect effect of Low State Antiquity are negative andstatistical significant in both the FE and the DID specifications. Furthermore, whencompared to the total effects reported at the bottom of the Table, the estimates suggestthat being in a low state capacity region accounts for something on the order of half ofthe increased probability of persecution due to negative temperature shocks. Theestimated effect size of 2.34 in the DID specification suggests in states with low antiquitythe effect of cold temperature on the probability of a persecution was approximatelytwice as high as in the baseline. The coefficients on the direct effect of Low State Antiquityare positive, which suggests that regardless of temperature, Jewish communities in lowstate capacity regions were more likely to be persecuted.

Our conceptual framework suggests that it was strong and stable states, notnecessarily states with constraints on the executive or rules protecting capital that weremore likely to be able to commit to rent-seeking arrangements with Jewishcommunities in the medieval and early modern period. Therefore, we want todisentangle the effect of state capacity from the effects of constraints on the executiveor protection of property rights.

To do this, columns (5)–(8) of Table 2 test whether the Acemoglu et al. (2005)measures of constraint on executive and protection of capital predict persecutions. Were-code their constraints variables so as to be comparable to our Low State Antiquityvariable by turning them into dummy variables where countries with constraints belowthe sample average get a one and high constraint countries are zeroes. We theninterpolate these values between centuries and extract their values for each city(descriptive statistics are in online Appendix A). We measure (lack of) constraints onthe executive using Low Constraints. The variable measuring (lack of) protection ofcapital is Low Capital Protection.

The regression results showing the effect of Low Constraints on persecutionprobability are in columns (5) and (6). The fixed effects specification indicates thatcities located in regions with less constraint were less likely to persecute Jews. The

32 The values for this last measure are 1 point if over 50%, 0.75 points if between 25% and 50%, 0.5 pointsif between 10% and 25%, 0.3 points if less than 10%.

33 Descriptive statistics are in online Appendix A. As a robustness check, we also code our state antiquityvariable as a continuous variable. We report these regressions in online Appendix B.



interaction of Low Constraints with Temperaturei;t�1 is statistically insignificant. Underthe DID specification, the level of the Low Constraints variable has the wrong sign.

When we look at our measure of capital protection in columns (7) and (8), theresults are similar. The sign on the direct effect of Low Capital Production is negative inour fixed effects specification indicating that less constraint in an area is associatedwith less persecution. The sign on the interaction is positive though insignificant.Under the DID specification in column (8), the coefficient on Low Capital Protection isconsistent with zero. Overall, these regressions support our argument that it was statecapacity that determined whether a Jewish community would be persecuted duringtimes of economic crisis.

The empirical findings we report in Table 2 provide evidence in support of ourhypothesis that the effects of colder weather would be greater in weak states. They alsoprovide some albeit less robust evidence that the effects of colder weather were greaterin areas with lower quality soil. There is no evidence that constraints on the executiveand protection of capital were associated with protection of minorities in the medievalor early modern period. As we discuss below, these results are consistent with a body ofhistorical research that suggests that while, in the long-run, state capacity andconstraints on the executive are complements (Besley and Persson, 2011), in themedium-run, investments in state capacity often preceded the imposition ofconstraints on the executive (Dincecco, 2009; Johnson and Koyama, 2014a). In thenext Section, we investigate the robustness of these results.

4. Robustness

4.1. Results with Yearly Data

In the previous Section, our empirical results were all based on five-year periods. In thisSection, we explore how robust our results are to using yearly data.

Table 3 shows the results from estimating 1 using yearly data. Our main variable ofinterest is the average temperature for the years t�1 to t�5.

34 To maintain symmetry withthe analysis above, in theTables we refer to this variable asTemperaturei;t�1. In column (1),we report the specification with city fixed effects. Our coefficient of interest is negative,and implies that a series of one degree colder growing seasons are associated with a 0.46percentage points higher probability of persecution. The DID specification in column(2) is smaller than the FE coefficient and less precisely estimated (p-value = 0.174).35

Table 4 reports the results from estimating 2 testing the effect of our hypothesisedmechanisms using the yearly data. Wheat suitability does not have a precisely estimatedeffect on persecution probability when we employ yearly data (cols. 1–2). However, the

34 If the persecution is in 1541, for example, then we use temperature data in the five-year period of 1536to 1540 to explain it.

35 The coefficients using the one-year data are typically about a fifth of those using the five-year averagedata, though the interpretation of the effect size is similar. A typical coefficient from the five-year DIDregressions is �1.25. This means a one degree decrease in temperature during the previous five years leads toa 1.25% increase in persecution probability over the next five years (or about 0.25% each year). Thecoefficient of �0.272 in Column 3 of revised Table 3 means a 1 degree decrease in temperature during theprevious five years leads to about a quarter of a percentage increase in persecution probability in just the nextyear (as opposed the next five years).



overall effect of wheat suitability and temperature remains large. Our measure of stateantiquity remains robust. Our largest estimate for the effect of weather on persecutionsis in the FE regression in column (3).

4.2. The Iberian Expulsions

One potential objection to our results is that the expulsions of Jews from Spain in 1492and Portugal in 1497 represent a large proportion of the variation in the outcome. Thetraditional explanations for the timing of the Spanish expulsions of 1492 is thatFerdinand and Isabella were driven by nationalist and religious motives and that thetiming of the expulsions simply followed from the unification of Spain following theconquest of Granada.36 Revisionist accounts also emphasise the importance of localelites in demanding the expulsion in return for supporting the monarchy (Haliczer,1973). When we exclude the Iberian expulsions the size of our coefficients shrink byroughly a half. A one standard deviation decrease in temperature now increases thebaseline probability from 2% to 2.5%. Therefore in qualitative terms our results arerobust to exclusion of the Iberian expulsions.

4.3. Placebo Regressions

An assumption of our identification strategy is that the temperature data are randomand, thus, unrelated to unobservables that could potentially lead to bias in ourestimates. We perform a placebo test in order to test this assumption.

We re-estimate 1 replacing the value for a city’s current temperature with those fromprevious years and for future years. Figure 3 depicts the coefficients that we obtainfrom this regression using yearly data. It demonstrates that the coefficient on

Table 3

Baseline Results Yearly Data

Baseline effects of temperature on persecutions and expulsions

Persecutions Expulsions

(1) (2) (3) (4) (5) (6)

Temperaturei;t�1 �0.346*** �0.464*** �0.272 �0.272*** �0.360*** �0.150(0.104) (0.124) (0.174) (0.075) (0.0837) (0.130)

Controls Y Y Y Y Y YCity dummies N Y Y N Y YTime dummies N N Y N N Y

N 276,359 276,359 276,359 276,359 276,359 276,359F 4.682 5.726 2.444 4.758 6.148 1.340

p-values 0.005 0.002 0.129 0.005 0.001 0.256

Notes. *p < 0.10, **p < 0.05, ***p < 0.01. Standard errors clustered at the climate grid level in parentheses.Observations are at the city 9 one-year level between 1100 and 1799. Otherwise controls are the same as thoseemployed in Table 1.

36 For studies of the expulsion of Jews from Spain see Kamen (1988), Gerber (1992), Roth (1995).



Tab

le4

YearlyMechanism

Regressions

Dep

enden

tvariab

le:persecu

tions

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Tem

peraturei;t�

1�0

.292

**�0

.168

�0.276

**�0

.156

�0.479

***

�0.284

�0.577

***

�0.278

(0.114

)(0.16

8)(0.310

)(0.687

)(0.657

)(1.343

)(0.728

)(1.064

)Low

Wheat

9Tem

peraturei;t�

1�0

.432

*�0

.212

(0.227

)(0.136

)Low

StateAntiquity

0.17

4*0.48

1***

(0.395

)(0.455

)Low

StateAntiquity9

Tem

peraturei;t�

1

�0.485

**�0

.269

(0.868

)(0.554

)Low

Constraints

�0.300

**�0

.284

(0.130

)(0.200

)Low

Constraints9

Lag

Tem

peraturei;t�

1

0.17

10.04

5(0.201

)(0.156

)Low

Capital

Protection

�0.378

*�0

.010

(0.214

)(0.549

)Low

Capital

Protection9

Tem

peraturei;t�

1

0.19

00.01

0(0.178

)(0.205

)

Tem

perature;t�1

+Interaction

�0.724

***

�0.380

*�0

.761

***

�0.426

*�0

.308

***

�0.240

�0.387

***

�0.268

*(0.210

)(0.211

)(0.202

)(0.213

)(0.096

)(0.150

)(0.107

)(0.150

)

Controls

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

CityDummies

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Tim

edummies

No

Yes

No

Yes

No

Yes

No

Yes

N27

6,35

927

6,35

927

6,35

927

6,35

927

6,35

927

6,35

927

6,35

927

6,35

9F

9.98

81.79

74.94

58.61

19.24

00.94

54.61

61.08

7

p-values

0.00

00.18

40.00

20.00

10.00

00.43

20.00

20.37

1

Notes.*

p<0.10

,**p

<0.05

,***

p<0.01

.Standarderrorsclustered

attheclim

ategrid

leveln

paren

theses.O

bservationsareat

thecity9

one-year

levelb

etwee

n11

00an

d17

99.Allco

ntrolsan

dvariab

ledetailsareotherwisethesameas

inTab

le2.



temperature is significant and fairly large up to about 5 lags thereby justifying our useof Temperaturei;t�1 as our main explanatory variable. Importantly, it shows that theeffects of future temperature are indistinguishable from zero.

4.4. Spatial and Serial Correlation

Temperature is highly spatially correlated. To ensure that our results are not biaseddue to spatial spillovers, in Table B2 in online Appendix B, we replicate our baselineand mechanism regressions controlling for spatial and serial correlation using themethod suggested by Conley (2008) and implemented in Stata code by Hsiang (2010).These results corroborate our main analysis.

4.5 Rainfall and Persecution

As a check on the robustness of our claim that supply shocks caused Jewishpersecution, we also compile data on rainfall between 1500 and 1799 to re-estimate ourmain regressions. For dry soil farming, wheat requires at least 60 cm of annual rainfall.Higher levels of rainfall are particularly important for increasing yields during thegrowing season as the the demand for water increases with leaf growth which peaksbetween July and August.37 Estimates for historical levels of rainfall at a disaggregated

–1.0

–0.5

0

0.5E

stim

ated

Eff

ect o

f T

empe

ratu

re o

n Pe

rsec

utio

n

–10 –9 –8 –7 –6 –5 –4 –3 –2 –1 0 1 2 3 4 5

Size of Lag/Lead of Temperature

Coefficient on Temperature 10% CI

Fig. 3. Placebo Regressions Using One-year DataNotes. Dotted lines are 10% confidence intervals. Vertical line marks t = 0.

37 Abnormally low summer temperatures are often associated with excess rainfall and this combinationwould be expected to lower yields. In general, however, too little rainfall was much more likely to reduceagricultural output.



level only exist from 1500 onwards therefore we do not include these data in our mainregressions. In online Appendix B, Tables B3 and B4, we report both our baseline andmechanism regressions for the period 1500–1799 using data from Pauling et al.(2006). Table B3 provides evidence that too little rainfall during the growing seasonresulted in a higher probability of persecution. The OLS regression in Column 1suggests a 10% decrease in rainfall raises the probability of a persecution in thesubsequent five-year period by 2%.38 In Column (3) where we report the OLS effectwith an interaction for the period before 1600, the overall effect increases such that a10% decrease in rain raises persecution probability by 5.8% Under the DIDspecifications in Columns (2) and (4), the effect of decreased rain is still present,though statistically insignificant. However, when we allow the city fixed effects to bedifferent before and after 1600, then effect is again robustly negative with a 10%reduction in rain leading to a 7% increase in persecution probability. Overall, therainfall regressions support our main finding that negative supply shocks inducedJewish persecutions in the early-modern period. The fact that the rainfall data aregenerated from a completely different source from the temperature series, we use andwe can show effects even though there are only 100 years of data during the period, webelieve supply shocks mattered most for Jewish communities (pre-1600) is encourag-ing support for our hypothesis.

Overall, our findings identify two forces that appear to have played a significant rolein persecution probability. First, temperature shocks appear to have had greater effectin locales with relatively unproductive agriculture. Second, Jewish communitieslocated in relatively weak polities were more likely to be persecuted during colderperiods. This provides evidence in favour of our argument that in the face of negativeeconomic shocks, weak rulers were less able to uphold the extractive rent-seekingarrangements that protected Jewish communities. In the next Section, we discussadditional evidence supportive of this explanation as well as assessing other hypothesesthat may explain why Jewish persecutions declined during the later early-modernperiod.

5. Discussion: The Decline of Jewish Persecutions

We have documented a robust relationship between periods of colder temperatureand Jewish persecutions. After around 1600, however, the relationship betweencolder temperature and the likelihood of persecution disappeared. This can beseen in Figure 4 which depicts the coefficients from our baseline regression(Table 1) estimated by century (the 12th century is the excluded category).39

38 The rainfall variable is constructed in a similar manner to the temperature variable in our baseline five-year regressions. The only difference is that we express it in logs.

39 We estimate the following regression:

yit ¼X18th

c¼13th

bcTi;t�1it � Ic þ gi þ lt þ X0itXþ eit; (3)

where Ic is a vector of dummies for each of the thirteenth to eighteenth centuries (the twelfth century is theexcluded category). Standard errors are clustered at the climate grid level.



The data suggest that Jewish persecutions were particularly intense in latemedieval and Renaissance Europe and often triggered by supply shocks. After1600, however, colder temperatures ceased to be associated with a greater risk ofpersecution.

In this Section, we discuss in greater detail five possible reasons for the decline inJewish persecutions:

(i) that there were simply fewer Jewish communities to persecute by theseventeenth century;

(ii) that improved agricultural productivity, or, better integrated markets couldhave reduced vulnerability to temperature shocks;

(iii) that the rise of stronger states could have led to more robust protection forreligious and ethnic minorities;

(iv) that there were fewer negative temperature shocks; and(v) that the impact of the Reformation and the Enlightenment may have reduced

antisemitic attitudes.

5.1. The Decline of Western and Central European Jewry?

The change in the geographic distribution of Jewish communities after 1500 maypartially account for the breakdown in the relationship between temperature shocksand Jewish persecutions. There was a definite move eastwards as Jewish communities,

–0.15

–0.10

–0.05

0

0.05

0.10C

oeff

icie

nt E

stim

ate

1250 1350 1450 1550 1650 1750

Century

Coefficient Estimate 95% CI

Fig. 4. The Effect of Temperature on Persecution Probability Over TimeNotes. Plots coefficients estimate by century of the baseline DID regression. Specificationdiscussed in the text.



persecuted first in Germany and then later in Spain and Portugal, fled to either easternEurope or the Ottoman Empire.40

The political equilibrium that governed the coexistence of Jews and Christiansdiffered in eastern Europe. Whereas in western Europe, Jews specialised asmoneylenders and came to be exploited as fiscal resources by medieval rulers, thisdid not occur to the same extent in medieval or early modern Poland. Historians notethat the condition of the Jews in Poland was therefore relatively favourable comparedto western Europe and remained so until the seventeenth century. Weinryb (1972, p. 51)notes that ‘Jews themselves, both within the country and abroad, looked upon PolishJews as living in security’. This reflected deeper institutional differences betweenPoland and the rest of Europe along a range of margins: Poland was also comparativelytolerant to heretics during the sixteenth century (Tazbir, 1973). One reason for thiswas that the economic underdevelopment of Poland meant that the possibilities foreconomic specialisation were also limited. As the division of labour was not as welldefined, Jews continued to play a number of important economic roles in the Polisheconomy throughout the medieval and early modern period; they were not restrictedto their role as moneylenders as was the case in western Europe after 1200 (Dubnow,1975).41 For this reason, it is likely that some of the migration to Poland may have beendriven by its reputation as a relatively safe haven.

In Table 5, we calculate the number of city-years that a Jewish community waspresent in various countries in our data before and after 1600. In the case of Spain andPortugal, the reason they stopped persecuting Jewish communities after 1600 isbecause they did not have any openly acknowledged communities after the nationalexpulsions of 1492 and 1497.42 However, Spain and Portugal are extreme cases. Thedata in Table 5 show that, while there was definitely a movement of Jewish

Table 5

City-years Before and After 1600

CountryCity-years

before 1600City-yearsafter 1600

Persecutionsbefore 1600

Persecutionsafter 1600

Persecutions per city yearbefore/after 1600

Austria 4,223 1,402 19 7 0.90England 3,530 1,414 55 0 ∞France 25,346 6,477 233 6 9.92Germany 41,080 19,254 335 24 6.54Italy 29,847 8,654 108 19 1.65Poland 16,012 21,814 47 66 0.97Portugal 5,734 0 21 0 ∞Spain 36,065 0 204 0 ∞Switzerland 2,890 697 25 2 3.02

Source. Encyclopedia Judaica (2007).

40 We do not discuss the Ottoman Empire, as the institutions governing the toleration of Jews in Ottomanterritories were very different to those in Christian Europe.

41 Consistent with this claim is the observation of Dubnow (1975) that there was antisemitic agitation inthe cities of western Poland as conditions there were more hospitable to Jews in the more rural parts ofcentral and eastern Poland.

42 The Inquisition conducted intense persecutions of conversos. We discuss this issue in online Appendix C.



communities eastward towards countries such as Poland after 1600, there were stillmany Jews in most of the countries in our sample. Furthermore, the last column ofTable 5 shows that, even if we express persecution probability in per city-year terms,there was a marked decline in Jewish persecution after 1600 for most countries.

5.2. Greater Agricultural Productivity or Increased Market Integration?

Another possible explanation for the reduction in the number of Jewish persecutionsin western Europe and the breakdown in the relationship between temperature andpersecutions is that an increase in agricultural productivity, or increased marketintegration, made European economies less vulnerable to supply shocks. This iscertainly one of the explanations that our conceptual framework suggests. From theeighteenth century onwards, Malthusian conditions weakened and per capita incomesgradually increased. Nunn and Qian (2011), for example, document the role played bythe potato in increasing population density and urbanisation after 1700. However, thefrequency of Jewish persecutions markedly declined from around 1600 – a periodwhen the European economy remained Malthusian and agricultural productivity waslow.43 Thus the Columbian Exchange occurred too late to explain the decline of Jewishpersecutions.

Nevertheless, better market integration might be responsible for temperature shockshaving a weaker impact on Jewish persecutions. To assess this, we use a panel of wheatprices from the Allen-Unger database.44 This data set contains grain prices for 193 citiesworldwide. We use wheat prices from the 98 European cities in this data set.45 In Table 6,we run a series of regressions of Temperature on the log of wheat prices in the cities in thisunbalanced panel. Columns (1) and (2) report our fixed effects andDID estimates usingthe entire panel. Becausewe expect a badharvest due to colder temperature to raise grainprices in the same year, we focus on contemporaneous grain prices.

These regressions support our hypothesis that negative temperature shocks adverselyaffect grain markets. The coefficient on Temperature in specification (1) implies a onedegree celsius decrease in temperature increases wheat prices by about 9%. Thisestimate is halved in the DID regression. In column (3), we include a lag of thedependent variable to account for potential serial correlation and the estimate shrinksfurther while retaining the negative sign and its significance.46 Temperature shocksaffected agricultural output as measured by grain prices.47

43 Recent research finds evidence that the Malthusian equilibrium weakened in England during theseventeenth century; see, for instance, Crafts and Mills (2009). But the overwhelming consensus is thatagricultural productivity remained low outside England and the Netherlands throughout the seventeenthcentury; see Allen (2000) amongst many others.

44 The Allen-Unger Global Commodities database is available at http://www.gcpdb.info/.45 Figure C8 in online Appendix C shows their locations. Descriptive statistics are in online Appendix A.46 A Fisher-type unit root test on the data convincingly rejects the null of a unit root. This implies the

inclusion of the lagged dependent variable term is unnecessary. Nonetheless, we include it as a cautionarymeasure.

47 While these coefficients are small, they are also consistent with what we know about medieval and earlymodern agriculture. Grain market prices were highly regulated and local governmental authoritiesprovisioned cities during periods of scarcity so adjustment to exogenous shocks was often through quantityrather than through price (for a discussion of the French case see Kaplan (2013)).



In columns (4)–(6), we assess whether temperature shocks affected grain pricesless after 1600 than before by interacting temperature with a dummy variable for allperiods before 1600. The fixed effects specification implies that a one degreecentigrade decrease in average temperature is associated with a 24% increase ingrain prices (column 4) before 1600. This suggests that increased marketintegration after 1600 reduced vulnerability to climatic shocks. Similar to theregressions on the whole sample, when we include time dummies and a lag of thedependent variable, these effect sizes shrink considerably. Overall, these resultssuggest that better markets and improvements in agricultural technology did meanthat European economies were less vulnerable to climate shocks after 1600.Nevertheless, while the effect of climate on agricultural prices weakened after 1600,it did not disappear. Therefore, improvements in market integration are unlikely tobe solely responsible for the marked decline in Jewish persecutions that occurred inthe early modern period.

5.3. Greater State Capacity?

Our conceptual framework suggests that in societies with greater state capacity therelationship between temperature shocks and Jewish persecutions should be weaker.We also find significant support for this proposition from the results we obtain usingour Low State Antiquity variables. These regressions indicate that the link betweenpersecution and colder temperatures was strongest in states of more recent origin.Unfortunately, it is difficult to test this hypothesis more directly since reliable andcontinuous data on tax revenues and other measures of state capacity are only availablefor the major European states after 1500 (and in some cases 1650) (Dincecco, 2009;

Table 6

Grain Price Regressions

Dependent variable: wheat prices (log)

(1) (2) (3) (4) (5) (6)

Temperature �8.96*** �4.43*** �1.68** �6.51*** �1.81 �0.89(1.127) (1.520) (0.817) (1.369) (1.598) (0.944)

Pre-1600 9Temperature

�18.34*** �7.866*** �2.451(2.834) (6.190) (2.791)

Pre-1600 �59.11*** �164.4*** �68.78***(6.530) (7.225) (4.297)

Temperature +Interaction

�24.86*** �9.67* �3.344(1.996) (5.202) (2.333)

Lag Grain Prices No No Yes No No YesMarket dummies Yes Yes Yes Yes Yes YesTime dummies No Yes Yes No Yes Yes

N 16,171 16,171 15,193 16,171 16,171 15,193Adjusted R2 0.008 0.594 0.771 0.219 0.594 0.771

Notes. *p < 0.05, **p < 0.01, ***p < 0.001. Observations are at the city 9 year level between 1100 and 1799.The dependent variable is the log of wheat prices in silver equivalents. Coefficients are multiplied by 100 andreported with standard errors clustered at the city level.



Readmission of Jews

0

5

10

15

Per

Cap

ita

Tax

Rev

enue

/Uns

kille

d D

aily

Wag

e

0

1000

2000

3000

Parl

iam

enta

ry A

cts

1500 1600 1700 1800Year

Parliamentary Acts Tax/Wage

0

20

40

60

80

No.

of

Deb

asem

ents

and

Edi

cts

2

4

6

8

10

12

Tax

Rev

enue

/Uns

kille

d D

aily

Wag

e

1400 1500 1600 1700 1800Year

Edicts

(b)

(a)

Tax Revenue/Wages Debasement

Fig. 5. The Rise of State Capacity in Early Modern Western Europe (a) England (b) France (c) DutchRepublic (d) Prussia

Notes. Tax revenue data is from Karaman and Pamuk (2013); data on English Parliamentary Actsis from Bogart and Richardson (2011). Data on Royal Edicts is from Johnson and Koyama(2014a); data on debasements is from Shaw (1896). Army size data is from van Nimwegen(2006).



1593 Arrival ofPortugese Jews

20,000

40,000

60,000

80,000

100,000

120,000

Arm

y Si

ze

10

15

20

25D

utch

Tax

es/W

ages

1500 1600 1700 1800Year

(c)

Dutch Taxes/Wages Army

1671 Jews Admitted into Brandenberg

0

50,000

100,000

150,000

200,000

Arm

y Si

ze

0

5

10

15

Prus

sian

Tax

es/W

ages

1500 1600 1700 1800Year

(d)

Prussian Taxes/Wages Army Size

Fig. 5. (Continued)



Karaman and Pamuk, 2013).48 The point at which systematic data becomes availablenot surprisingly coincides with the rise of strong centralised states. Furthermore,estimates of tax revenues generally do not exist for the many city-states and smallerpolitical units that comprise much of our data.

Significant narrative evidence supports our argument that increases in fiscal capacityin western Europe were linked to the decision to admit and protect Jewishcommunities. Figure 5a depicts the rise of state capacity in England as measured bytax revenue and the increase in legislative activity by Parliament. For example, OliverCromwell invited Jews to return to England in 1655 – a period during which statecapacity was expanding in England (Kaplan, 2007, p. 326). The English Jewishcommunity was initially viewed as a transient group of alien merchants with limitedrights. Their position was uncertain and in 1660 a petition came before Parliament toexpel them. This petition was ignored but it was only after the Glorious Revolutionthat the permanent status of Jews in England was fully recognised and accepted (Katz,1994, pp. 140–1, 188). Despite the unpopularity of the English Jewish community, theywere no longer subject to persecutions and violence after this official recognition.49

In the English case, it is difficult to disentangle the importance of state capacity inprotecting the Jews from persecution from the role Parliament may have played inprotecting them. However, this is not the case in France where a similar process tookplace under the auspices of an absolutist monarchy. Figure 5b traces the dramaticincrease in the capacity of the central state after around 1630, using tax revenues andcounts of royal edicts issued. Cardinal Richelieu played a crucial role in establishingthe French tax state and in protecting Portuguese crypto-Jews from being persecuted asheretics because of their value as merchants and financiers. Israel (1985, pp. 96–7)describes this as ‘a classic instance of raison d’ �Etat politics and mercantilism’ the resultof which was to have ‘made possible that steady transition from the 1630s down to the1680s by when the Portuguese communities in France had cast off all remainingpretence and openly organised as Jewish congregations with rabbis and services inHebrew’. By 1722, the right of all French Jews to openly practice their religion wasrecognised in law.

The Dutch Republic offered permanent protection to Jews after its declarationof independence from Spain, with large numbers of the so-called crypto-Jewsarriving in 1593. The rights of Jews to practice their religion was codified in 1619.Figure 5c plots the relationship between toleration offered to the Jews and the riseof the Dutch state as measured by taxation per capita over real wages and by thewartime strength of the Republic’s armed forces. Figure 5d depicts similar data forPrussia. While there had been a long history of Jewish settlement in landscontrolled by Prussia, it was only in the late seventeenth century that FrederickWilliam (1650–88) gave the Jews a charter, which established their permanentresidency. This occurred at the same time as the Elector invested in fiscal capacity

48 In online Appendix D, Figure D3 depicts the rise in tax revenues (measured in silver) for the seven largestEuropean states between 1500 and 1800.

49 For details on the survival of antisemitic stereotypes and attitudes in England after the re-admittance ofJews into the country see Poliakov (1955, pp. 203–9) and Felsenstein (1999).



and built a professional standing army that would propel Prussia to the status of amajor European power.

There can be no doubt that popular antisemitism survived the emergence ofstronger nation states in the early modern period.50 But the evidence suggests thatthese new states were less responsive to it. The increases in state capacity thatoccurred from 1600 onwards, documented by Dincecco (2009); Karaman and Pamuk(2013) and Johnson and Koyama (2014a), led to the formation of polities that wereless vulnerable to political unrest and better at reducing interfaith violence, allfactors that led to fewer persecutions and expulsions.51 This is consistent with thefindings that stronger states were responsible for ending the European witch-hunts inthe late seventeenth century (Levack, 1996; Johnson and Koyama, 2014b) and withthe argument that the rise of larger and more centralised states led to a gradualincrease in bounds of religious toleration in the early modern period (Johnson andKoyama, 2013).52

Jews continued to suffer persecutions and massacres in early modern Europe butthese occurred in the ungoverned periphery and not in the new nation states ofwestern Europe. The worst massacres occurred during the Khmelnytsky Uprising,which saw the Ukraine breakaway from Poland-Lithuania in the mid-seventeenthcentury (Stampfer, 2003). It goes without saying that this increase in state capacity wasa two-edged sword: it could be used to persecute as well as protect. In the twentiethcentury, the capacity of modern states made possible the industrial horrors of theHolocaust. But in the period between 1600 and 1800, it was associated with a reductionin violence against minority groups.

5.4. Colder Temperature?

The period 1400 to 1800 is known as the Little Ice age. This name is perhaps slightlymisleading as there was a lot of variation in climate within this long period. Forexample, Crowley and Lowery (2000) note that the medieval warm period comprisedthree distinct and temporal separate peaks in temperature: 1010–1040, 1070–1105, and1155–1190 and that average temperatures in the medieval warm period were only 0:2�Cwarmer than during the Little Ice Age. Consequently, some scholars have questioned

50 There is little evidence of a lessening in antisemitic attitudes. Judensau – woodcut images denigratingJews – remained common in Germany until 1800. Poliakov (1955, pp. 174–202) examines a large number ofantisemitic treatises published in France during the seventeenth century that suggest that antisemitism waswidespread and conventional in both elite and popular circles.

51 Historians and sociologists have also argued that the birth of new nation states in the late medievalperiod was often accompanied by the expulsion of the Jews and other ‘alien’ populations (Baron, 1967a;Menache, 1987; Barkey and Katznelson, 2011). However, this was not in general true of medievalpersecutions or expulsions. For studies of the expulsion of Jews from England see Leonard (1891), Elman(1937), Ovrut (1977), Menache (1987), Stacey (1997, 2000), Mundill (1998), Katznelson (2005), Koyama(2010b).

52 According to Heckscher (1955, p. 305): ‘The same tendency is manifested in the fact that the Jews wereplaced on a new footing in the seventeenth century in most western and central European countries. Thisshould certainly not be regarded as a general pro-Jewish feeling on the part of mercantilists. No suchsentiment was ever felt among those in power . . . this much is clear, that the leaders of mercantilist policywished to extend toleration even to the Jews, and that this toleration was determined primarily by commercialconsiderations’.



the value of terms such as the ‘medieval warm period’ or the ‘Little Ice Age’ becausewithin period variation is often much larger than between-period variation (Kelly andGr�ada, 2014).53

In our data, average temperatures in Europe between 1400 and 1600 were between0.10 and 0.20 degrees cooler than during the surrounding centuries. Nevertheless,even though the decline in mean temperature was modest, the period after 1400 sawgreater temperature volatility. In particular, ‘bad weather – heavy rainfall, cool and wetsummers, severe floods – became much more frequent during the last two centuries ofthe Middle Ages and suggest that, at least in north-western Europe, climatic disasterswere in fact more numerous and severe than they had been earlier. This is supportedby other evidence: the extension of the glaciers in the Alps; the lowering of the upwardlimits of cultivation and of tree growth in hilly areas; the inundation of the coastallowlands in the Netherlands and the increasing soil moisture in the valley of centralEurope, where, in some instances, cultivation had to be abandoned’ (Pounds, 1974, p.136).

Does an improvement in temperature account for the decline in persecutions? Inour data, summer growing conditions improved in the seventeenth century but the lateseventeenth century again saw extremely cold winters due to a decline in solar activityknown as the Maunder Minimum. Indeed, many scholars have argued that coldweather and particularly low winter temperatures during this period were associatedwith the political turbulence known as the crisis of the seventeenth century (Parker,2013). Certainly, a sustained improvement in climatic conditions did not begin untilthe eighteenth century. Thus, it is unlikely that improving climatic conditions canexplain the decline in Jewish persecutions.

5.5. Changes in Values and Beliefs?

A final factor we do not consider in detail is the importance of changing cultural valuesand beliefs, as emphasised by Mokyr (2002, 2009). The Reformation did not lead to amarked decline in religious tension or antisemitism. Erasmus, the leading advocate ofChristian humanism, is viewed by historians as an advocate of greater religioustoleration, at least for Trinitarian Christians (Zagorin, 2003, pp. 50–63). However, hedid not favour religious pluralism and ‘would have had no patience with the modern,enlightened idea of of toleration – of individual rights that extend to every race andcreed’ (Oberman, 1981, p. 39). In particular, his lenient attitude towards otherChristians did not mean that he favoured extending better treatment to Jews for whomhe had an ‘unbound hatred’ suspecting them of a ‘collective conspiracy’ and viewingthem as ‘culpable as the wirepullers of the German Peasant’s War (Oberman, 1981,p. 38).54 Martin Luther expressed increasingly antisemitic views once he realised thatthe Reformation would not win over large numbers of Jewish converts; he penned the

53 Kelly and Gr�ada (2014) contend that there was no decrease in temperature during the late middle ages.However, their argument remains contested by climate scientists and historians (B€untgen and Hellmann,2014; White, 2014).

54 Whether Erasmus’s anti-Judaism is sufficiently virulent to represent antisemitism is a subject of scholarlydebate (Markish, 1986).



book The Jews and their Lies in 1543 and railed against them for usury: ‘they [the Jews]are nothing but thieves and robbers who daily eat no morsel and wear no thread ofclothing which they have not stolen and pilfered from us by means of their accursedusury’. He was also an advocate of expulsion: ‘eject them forever from this country.For, as we have heard, God’s anger with them is so intense that gentle mercy will onlytend to make them worse and worse, while sharp mercy will reform them but little’(Luther, 1553 [1971], Chapter 15).

From the end of the seventeenth century onwards, however, the Enlightenment mayhave played a role in reducing antipathy towards Jews at least among elites (Kamen,1967; Grell and Porter, 2000). Data do not exist that shed light on the role played bythe Enlightenment in increasing religious toleration. Certainly, by de-emphasising theimportance of revealed religion, the Enlightenment created a religiously neutralsphere where Christians and Jews could meet on an equal footing (Low, 1979).Nevertheless many Enlightenment Figures including Voltaire, Edward Gibbon andEdmund Burke voiced anti-Jewish views (Sutcliffe, 2000). John Locke was in favour ofgranting toleration to Jews but this was because he believed it would encourage theirconversion to Christianity (Matar, 1993). Voltaire described the Jewish nation as ‘themost detestable ever to have sullied the earth’ (Sutcliffe, 2000, p. 32). It was only in thelate eighteenth century that Enlightenment views began to exert a decisive influenceon the treatment of Jews in Europe, generating the move towards Jewish emancipation.This intellectual movement, shaped by Moses Mendelssohn (1729–86), GottholdLessing (1729–81) and Christian von Dohm (1751–1820), and many others, came toolate to help explain the end of large-scale Jewish persecutions in western and centralEurope.55 Moreover, when Jewish emancipation did arrive at the end of the eighteenthcentury, it was shaped as much by the desire of rulers to make better economic use oftheir Jewish populations as it was by Enlightenment philosophy.56

6. Conclusion

This article examines the effect of negative supply shocks on the treatment of religiousor ethnic minorities. We develop a simple conceptual framework to study the conditionsunder which rulers find it beneficial to expel or expropriate a minority community. Inour empirical analysis, we exploit the fact that the economies of medieval and earlymodern Europe were predominantly agrarian and use exogenous variation intemperature during the growing season to identity the effect of supply shocks on theprobability of a Jewish community suffering persecution. A one standard deviationdecrease in average temperature increased the probability of a Jewish community beingpersecuted from a baseline of 2% every five years to between 2.5% and 3%.

55 Mendelssohn, however, did play a role in preventing an expulsion of Jews from Endlingen and Lengnauby writing to Johann Kaspar Lavater in 1774 (Hochman, 2014).

56 The Habsburg emperor Joseph II began the process of granting Jews civic rights in 1782. But it was theFrench Revolution and the subsequent invasion of Germany by French armies that led to the imposition ofJewish emancipation in central Europe (Berkovitz, 1989; Vital, 1999). After the defeat of France, thesereforms were partially reversed but the movement towards Jewish emancipation resumed and culminatedwith the removal of all disabilities on Jews in Austria-Hungary in 1868 and Germany in 1870 (Katz, 1974;Mahler, 1985; Sorkin, 1987).



We argue that more developed states with greater fiscal capacity and greater politicalstability were less likely to expel Jewish communities as a result of periods of coldtemperature. Our results support this hypothesis: the effect of supply shocks onpersecutions was greater in societies with lower state capacity. Persecutions peaked inthe fourteenth to sixteenth centuries. Increased agricultural productivity, greatermarket integration and the rise of centralised states helped account for Europe’sgradual transition from extractive to inclusive economic and political institutionsduring this period and in the establishment of a degree of protection for religiousminorities.

University of Michigan–DearbornGeorge Mason University

Submitted: 30 December 2013Accepted: 2 July 2015

Additional Supporting Information may be found in the online version of this article:

Appendix A. Descriptive Statistics.Appendix B. Further Robustness Tests.Appendix C. Data.Appendix D. Additional Historical Evidence and Discussion.Data S1.

ReferencesAcemoglu, D. (2003). ‘Why not a political Coase Theorem? Social conflict, commitment and politics’, Journal

of Comparative Economics, vol. 31(4), pp. 620–52.Acemoglu, D., Hassan, T.A. and Robinson, J.A. (2011). ‘Social structure and development: a legacy of the

Holocaust in Russia’, Quarterly Journal of Economics, vol. 126(2), pp. 895–946.Acemoglu, D., Johnson, S. and Robinson, J. (2005). ‘The rise of Europe: Atlantic trade, institutional change,

and economic growth’, American Economic Review, vol. 95(3), pp. 546–79.Acemoglu, D. and Robinson, J.A. (2012). Why Nations Fail, New York: Crown Business.Allen, R.C. (2000). ‘Economic structure and agricultural productivity in Europe, 1300–1800’, European Review

of Economic History, vol. 4(01), pp. 1–25.Anderson, R.W. (2012). ‘Measuring the Portuguese Inquisition’, Memo, University of Michigan-Dearborn.Bai, Y. and Kai-sing Kung, J. (2011). ‘Climate shocks and Sino-nomadic conflict’, The Review of Economics and

Statistics, vol. 93(3), pp. 970–81.Barber, M. (1981a). ‘Lepers, Jews and Moslems: the plot to overthrow Christendom in 1321’, History, vol. 66

(216), 1–17.Barber, M. (1981b). ‘The Pastoureaux of 1320’, Journal of Ecclesiastical History, vol. 32(2), pp. 144–66.Barkey, K. and Katznelson, I. (2011). ‘States, regimes, and decisions: why Jews were expelled from medieval

England and France’, Theory and Society, vol. 40(5), pp. 475–503.Baron, S.W. (1965a). A Social and Religious History of the Jews, Volume IX: Under Church and Empire, New York:

Columbia University Press.Baron, S.W. (1965b). A Social and Religious History of the Jews, Volume X: On the Empire’s Periphery, New York:

Columbia University Press.Baron, S.W. (1967a). A Social and Religious History of the Jews, Volume XI: Citizen or Alien Conjurer, New York:

Columbia University Press.Baron, S.W. (1967b). A Social and Religious History of the Jews, Volume XII: Economic Catalyst, New York: Columbia

University Press.Baron, S.W. (1975). Economic History of the Jews, Jerusalem: Keter Publishing House (edited by Nachum Gross).Barzel, Y. (1992). ‘Confiscation by the ruler: the rise and fall of Jewish lending in the Middle Ages’, Journal of

Law and Economics, vol. 35(1), pp. 1–13.



Berkovitz, J.A. (1989). The Shaping of Jewish Identity in Nineteenth-century France, Detroit, MI: Wayne StateUniversity Press.

Besley, T. and Persson, T. (2011). Pillars of Prosperity, Princeton, NJ: Princeton University Press.Bockstette, V., Chanda, A. and Putterman, L. (2002). ‘States and markets: the advantage of an early start’,

Journal of Economic Growth, vol. 7(4), pp. 347–69.Bogart, D. and Richardson, G. (2011). ‘Property rights and parliament in industrializing Britain’, Journal of

Law and Economics, vol. 54(2), pp. 241–74.Bosker, M., Buringh, E. and van Zanden, J.L. (2013). ‘From Baghdad to London: unravelling urban

development in Europe and the Arab world 800–1800’, Review of Economics and Statistics, vol. 95(4),pp. 1418–37.

Botticini, M. (1997). ‘New evidence on Jews in Tuscany, CA. 1310–1435: the ‘friends & family’ connectionagain’, Zakhor. Rivista di Storia degli Ebrei d’Italia, vol. 1, pp. 77–93.

Botticini, M. (2000). ‘A tale of ‘benevolent’ governments: private credit markets, public finance, and the roleof Jewish lenders in Medieval and Renaissance Italy’, Journal of Economic History, vol. 60(1), pp. 164–89.

Botticini, M. and Eckstein, Z. (2012). The Chosen Few, Princeton, New Jersey: Princeton University Press.Box-Steffensmeier, J. and Jones, B. (2004). Event History Analysis: A Guide for Social Scientists, Cambridge:

Cambridge University Press.Brown, R.B. and McCartney, S. (2005). ‘The exchequer of the Jews revisited: the operation and effect of the

Saccarium Judeorum’, Medieval History Journal, vol. 8(2), pp. 303–22.B€untgen, U. and Hellmann, L. (2014). ‘The little ice age in scientific perspective: cold spells and caveats’,

Journal of Interdisciplinary History, vol. 44(3), pp. 353–68.Campbell, B.M. (2010). ‘Nature as historical protagonist’, Economic History Review, vol. 63(2), pp. 281–314.Chaney, E. (2013). ‘Revolt on the Nile: economic shocks, religion and political power’, Econometrica vol. 81

(5), pp. 2033–53.Chazan, R. (1973–74). ‘Anti-usury efforts in thirteenth century Narbonne and the Jewish response’,

Proceedings of the American Academy for Jewish Research, vol. 41, pp. 45–67.Chazan, R. (2006). The Jews of Western Christendom, Cambridge: Cambridge University Press.Chazan, R. (2010). Reassessing Jewish Life in Medieval Europe, Cambridge: Cambridge University Press.Chua, A. (2004). World on Fire: How Exporting Free Market Democracy Breeds Ethnic Hatred and Global Instability,

New York: Anchor Books.Conley, T.G. (2008). ‘Spatial econometrics’, in (S.N. Durlauf and L.E. Blume, eds.), The New Palgrave

Dictionary of Economics, pp. 741–7, Basingstoke: Palgrave Macmillan.Crafts, N. and Mills, T.C. (2009). ‘From Malthus to Solow: how did the Malthusian economy really evolve?’,

Journal of Macroeconomics, vol. 31(1), pp. 68–93.Cramer, A.C. (1940). ‘The Jewish Exchequer, an inquiry into its fiscal function’, American Historical Review,

vol. 45(2), pp. 327–37.Crowley, T.J. and Lowery, T.S. (2000). ‘How warm was the medieval warm period?’, Ambio, vol. 29(1), pp. 51–

4.De Long, J.B. and Shleifer, A. (1993). ‘Princes and merchants: European city growth before the industrial

revolution’, Journal of Law & Economics, vol. 36(2), pp. 671–702.Dell, M., Jones, B. and Olken, B. (2012). ‘Temperature shocks and economic growth: evidence from the last

half century’, American Economic Journal: Macroeconomics, vol. 4(3), pp. 66–95.Dell, M., Jones, B.F. and Olken, B.A. (2014). ‘What do we learn from the weather? The new climate-economy

literature’, Journal of Economic Literature, vol. 53(3), pp. 740–98.Dincecco, M. (2009). Fiscal centralization, limited government, and public revenues in Europe, 1650–1913’,

Journal of Economic History, vol. 69(1), 48–103.Dobson, B. (2003). ‘The medieval York Jewry reconsidered’, in ( P. Skinner, ed.), The Jews in Medieval Britain:

Historical, Literary, and Archaeological Perspectives, pp. 145–56, New York: Boydell and Brewer.Dubnow, S.M. (1975). History of the Jews in Russia and Poland, Volume I. Ktav Publishing House (translated by

I. Friedlaender).Durante, R. (2009). Risk, cooperation and the economic origins of social trust: an empirical investigation.

MPRA Paper 25887, University Library of Munich.Elman, P. (1937). ‘The economic causes of the expulsion of the Jews in 1290’, Economic History Review,

vol. 7(2), pp. 145–54.Emery, R.W. (1959). The Jews of Perpignan in the Thirteenth Century: An Economic Study Based on Notoral Records,

New York: Columbia University Press.Encyclopedia Judaica (2007). 2nd edn, Detroit, MI: Thomson-Gale.Felsenstein, F. (1999). Anti-Semitic Stereotypes: A Paradigm of Otherness in English Popular Culture, 1660–1830,

Baltimore, MD: Johns Hopkins University Press.Fenske, J. and Kala, N. (2013). ‘Climate, ecosystem resilience and the slave trade’, CEPR Discussion Paper

no. 9449.



Fischer, G., van Nelthuizen, H., Shah, M. and Nachtergaele, F. (2002). Global Agro-ecological Assessment forAgriculture in the 21st Century: Methodology and Results, Rome: Food and Agriculture Organization of theUnited Nations.

Galloway, P.R. (1986). ‘Long-term fluctuations in climate and population in the preindustrial era’, Populationand Development Review, vol. 12(1), pp. 1–24.

Gerber, J.S. (1992). The Jews of Spain: A History of the Sephardic Experience, New York: The Free Press.Glaeser, E.L. (2005). ‘The political economy of hatred’, Quarterly Journal of Economics, vol. 120(1), pp. 45–86.Greene, W. (2004). ‘The behaviour of the maximum likelihood estimator of limited dependent variable

models in the presence of fixed effects’, Econometrics Journal, vol. 7(1), pp. 98–119.Grell, O.P. and Porter, R. (2000). ‘Toleration in enlightenment Europe’, in (O.P. Grell and R. Porter, eds.),

Toleration in Enlightenment Europe, pp. 1–23, Cambridge: Cambridge University Press.Grosfeld, I., Rodnyansky, A. and Zhuravskaya, E. (2013). ‘Persistent anti-market culture: a legacy of the Pale

of Settlement after the Holocaust’, American Economic Journal: Economic Policy, vol. 5(3), pp. 189–226.Guiot, J. and Corona, C. (2010). ‘Growing season temperature in Europe and climate forcings over the past

1400 years’, Plos One, vol. 5(4), pp. 1–15.Haliczer, S.H. (1973). ‘The Castilian urban patriciate and the Jewish expulsions of 1480–92’, American

Historical Review, vol. 78(1), pp. 35–62.Heckscher, E.F. (1955). Mercantilism, Volume II. London: George Allen & Unwin (translated by

E.F. Soderlund).Hochman, L. (2014). The Ugliness of Moses Mendelssohn: Aesthetics, Religion, and Morality in the Eighteenth Century,

London: Routledge.Horowitz, D. (2001). The Deadly Ethnic Riot, Berkeley, CA: University of California Press.Hsiang, S.M. (2010). ‘Temperatures and cyclones strongly associated with economic production in the

Caribbean and Central America’, Proceedings of the National Academy of Sciences, vol. 107(35), pp. 15367–72.Hsiang, S. and Burke, M. (2014). ‘Climate, conflict, and social stability: what does the evidence say?’, Climate

Change, vol. 123(1), pp. 39–55.Hsiang, S., Burke, M. and Miguel E. (2013). Quantifying the influence of climate on human conflict. Science,

vol. 341(6151), doi: 10.1126/science.1235367.Israel, J. (1985). European Jewry in the Age of Mercantilism, 1550–1750, Oxford: Oxford University Press.Jia, R. (2014). ‘Weather shocks, sweet potatoes and peasant revolts in historical China’, ECONOMIC JOURNAL, vol.

124(575), pp. 92–118.Johnson, N.D. and Koyama, M. (2013). ‘Legal centralization and the birth of the secular state’, Journal of

Comparative Economics, vol. 41(4), pp. 959–78.Johnson, N.D. and Koyama, M. (2014a). ‘Tax farming and the origins of state capacity in England and

France’, Explorations in Economic History, vol. 51(1), pp. 1–20.Johnson, N.D. and Koyama, M. (2014b). ‘Taxes, lawyers, and the decline of witch trials in France’, Journal of

Law and Economics, vol. 57(1), pp. 77–112.Jones, B. and Olken, B. (2010). ‘Climate shocks and exports’, American Economic Review: Papers and Proceedings,

vol. 100(2), pp. 454–9.Jordan, W.C. (1989). The French Monarchy and the Jews: From Philip Augustus to the Last Capetians, Philadelphia,

PA: University of Pennsylvania Press.Jordan, W.C. (1998). ‘Jews, regalian rights, and the constitution in Medieval France’, AJS Review, vol. 23(1),

pp. 1–16.Kamen, H. (1967). The Rise of Toleration, New York: World University Library.Kamen, H. (1988). ‘The Mediterranean and the expulsion of Spanish Jews in 1492’, Past & Present, vol. 119

(May), pp. 30–55.Kaplan, B. (2007). Divided by Faith, Cambridge, MA: Harvard University Press.Kaplan, S. (2013). Bread, Politics and Political Economy in the Reign of Louis XV: Anthem European Studies. Anthem

Other Canon Economics Series, London: Anthem Press.Karaman, K. and Pamuk, S. (2013). ‘Different paths to the modern state in Europe: the interaction between

warfare, economic structure and political regime’, American Political Science Review, vol. 107(3), pp. 603–26.

Katz, J. (1974). Out of the Ghetto, Cambridge, MA.: Harvard University Press.Katz, D.S. (1994). The Jews in the History of England, 1485–1850, Oxford: Clarendon Press.Katznelson, I. (2005). ‘“To give counsel and to consent”: Why the king Edward I expelled his Jews (in 1290)’,

in (I. Katznelson and B.R. Weingast, eds.), Preferences and Situations, pp. 88–126, New York: Russell SageFoundation.

Kelly, M. and Gr�ada, C. �O. (2014). ‘The waning of the little ice age: climate change in early modern Europe’,Journal of Interdisciplinary History, vol. 44(3), pp. 301–25.

Koyama, M. (2010a). ‘Evading the “taint of usury”: the usury prohibition as a barrier to entry’, Explorations inEconomic History, vol. 47(4), pp. 420–42.

Koyama, M. (2010b). ‘The political economy of expulsion: the regulation of Jewish moneylending inmedieval England’, Constitutional Political Economy, vol. 21(4), pp. 374–406.



Langmuir, G.I. (1990). Toward a Definition of Antisemitism, Berkeley, CA: University of California Press.Leonard, G.H. (1891). ‘The expulsion of the Jews by Edward I an essay in explanation of the Exodus, A.D.

1290’, Transactions of the Royal Historical Society, vol. 5(December), pp. 103–46.Levack, B.P. (1996). ‘State-building and witch hunting in early modern Europe’, in (J. Barry, M. Hester, and

G. Roberts, eds.), Witchcraft in Early Modern Europe, pp. 96–118, Cambridge: Cambridge University Press.Lipman, V.D. (1967). The Jews of Medieval Norwich, London: Jewish Historical Society of England (edited by

A.M. Habermann), with an Appendix of Latin documents from the Westminster Abbey MunimentRoom; and the Hebrew Poems of Meir of Norwich (edited by A.M. Habermann).

Low, A.F. (1979). Jews in the Eyes of Germans: from the Enlightenment to Imperial Germany. Michigan, MI: Institutefor the study of Human Issues.

Luther, M. (1553) [1971]. ‘On the Jews and their lies’, in (F. Sherman, ed.), Luther’s Works, Vol. 47: TheChristian in Society IV, pp. 121–306. Minneapolis, MN: Fortress Press.

Maccini, S. and Yang, D. (2009). ‘Under the weather: health, schooling, and economic consequences of early-life rainfall’, American Economic Review, vol. 99(3), pp. 1006–26.

Madesam, A., Shoag, D., Veuger, S. and Yanagizawa-Drott, D. (2013). ‘Do political protests matter? Evidencefrom the tea party movement’, Quarterly Journal of Economics, vol. 128(4), pp. 16633–85.

Mahler, R. (1985). Hasidism and the Jewish Enlightenment, Philadelphia, PA: The Jewish Publication Societyof America. translated from Yddish by Eugene Orenstein, translated from Hebrew by Aaron Klein andJenny Machlowitz Klein.

Malik, A. and Temple, J.R. (2009). ‘The geography of output volatility’, Journal of Development Economics, vol.90(2), pp. 163–78.

Markish, S. (1986). Erasmus and the Jews, Chicago, IL: University of Chicago Press.Masschaele, J. (1993). ‘Transport costs in medieval England’, Economic History Review, vol. 46(2), pp. 266–79.Matar, N.I. (1993). ‘John Locke and the Jews’, Journal of Ecclesiastical History, vol. 44(1), 45–62.Menache, S. (1985). ‘Faith, myth, and politics: the sterotype of the Jews and their expulsion from England

and France’, Jewish Quarterly Review, vol. 75(4), pp. 351–74.Menache, S. (1987). ‘The king, the Church and the Jews: some considerations on the expulsions from

England and France’, Journal of Medieval History, vol. 13(3), pp. 223–36.Menache, S. (1997). ‘Mathew Paris’s attitudes toward Anglo-Jewry’, Journal of Medieval History, vol. 23(2),

pp. 139–62.Miguel, E. (2005). ‘Poverty and witch killing’, Review of Economic Studies, vol. 72(4), pp. 1153–72.Mitra, A. and Ray, D. (2013). Implications of an economic theory of conflict: Hindu-Muslim violence in India.

NBER Working Paper 19090, National Bureau of Economic Research, Inc.Mokyr, J. (2002). The Gift of Athena: Historical Origins of the Knowledge Economy, Princeton, NJ: Princeton UP.Mokyr, J. (2009). The Enlightened Economy, New Haven, CT: Yale University Press.Moore, R.I. (1992). ‘Anti-Semitism and the Birth of Europe’, in (D. Wood, ed.) Christianity and Judaism,

Studies in Church History, vol. 29, pp. 33–57, Woodbridge, UK: Boydell & Brewer.Moore, R.I. (2008). ‘The war against heresy in medieval Europe’, Historical Research, vol. 81(212), pp. 189–

210.Mundill, R.R. (1991). ‘Lumbard and Son: the businesses and debtors of two Jewish moneylenders in late

thirteenth-century England’, Jewish Quarterly Review, vol. 82, pp. 137–70.Mundill, R.R. (1998). England’s Jewish Solution: Experiment and Expulsion, 1262–1290. Cambridge: Cambridge

University Press.Mundill, R.R. (2010). The King’s Jews: Money, Massacre and Exodus in Medieval England, London: Continuum.van Nimwegen, O. (2006). The Dutch Army and the Military Revolutions, 1599-1688, Woodbridge: The Boydell

Press (translated by Andrew May).Nirenberg, D. (1996). Communities of Violence, Princeton, NJ: Princeton University Press.North, D.C., Wallis, J.J. and Weingast, B.R. (2009). Violence and Social Orders: A Conceptual Framework for

Interpreting Recorded Human History, Cambridge: Cambridge University Press.Nunn, N. and Qian, N. (2011). ‘The potato’s contribution to population and urbanization: evidence from a

historical experiment’, Quarterly Journal of Economics, vol. 126(2), pp. 593–650.Oberman, H.A. (1981). The Roots of Anti-Semitism. Philadelphia, PA: Fortress Press. translated by James

I. Porter.Oster, E. (2004). ‘Witchcraft, weather and economic growth in Renaissance Europe’, Journal of Economic

Perspectives, vol. 18(1), pp. 215–28.Ovrut, B.D. (1977). ‘Edward I and the expulsion of the Jews’, Jewish Quarterly Review, vol. 67, pp. 224–35.Parker, G. (2013). Global Crises: War, Climate Change and Castastrophe in the Seventeenth Century, New Haven, CT:

Yale University Press.Pascali, L. (2015). Banks and development: Jewish communities in the Italian Renassiance and current

economic performance. Review of Economics and Statistics, forthcoming.Pauling, A., Luteracher, J., Casty, C. and Wanner H. (2006). ‘Five hundred years of gridded high-resolution

precipitation reconstructions over europe and the connection to large-scale circulation’, ClimateDynamics, vol. 26(4), pp. 387–405.



Pinker, S. (2011). The Better Angels of Our Nature: Why Violence has Declined, London: Viking.Poliakov, L. (1955). The History of Anti-Semitism, Volume 1. Philadelphia, PA: University of Pennsylvania Press

(translated by Richard Howard).Porter, J.R. and Gawith, M. (1998). ‘Temperatures and the growth and development of wheat: a review’,

European Journal of Agronomy, vol. 10(1), pp. 23–36.Porter, J.R. and Semenov, M.A. (2005). ‘Crop responses to climatic variation’, Philosophical Transactions of the

Royal Society, vol. 360, pp. 2021–35.Pounds, N. (1974). An Economic History of Medieval Europe, London: Longman.Richardson, H. (1960). The English Jewry under Angevin Kings, London: Methuen.Roth, C. (1961). ‘The economic history of the Jews’, Economic History Review, vol. 14(1), pp. 131–5.Roth, N. (1995). Conversos, Inquisition, and the Expulsion of the Jews from Spain. Madison, WI: The University of

Wisconsin Press.Shaw, W. (1896). The History of Currency, 1252 to 1894, 2nd edn, New York: Burt Franklin.Slavin, P. (2012). ‘The Great Bovine Pestilence and its economic and environmental consequences in

England and Wales, 1318–50’, Economic History Review, vol. 65(4), pp. 1239–66.Sorkin, D. (1987). ‘The genesis of the ideology of emancipation: 1806–1840’, Leo Baeck Institute Yearbook 32

(1), pp. 11–40.Stacey, R.C. (1985). ‘Royal taxation and the social structure of medieval Anglo-Jewry: the tallages of 1239-

1242’, Hebrew Union College Annual, vol. 56, pp. 175–249.Stacey, R.C. (1995). ‘Jewish lending and the medieval English economy’, in (R.H. Britnell and B.M.

Campbell, eds.), A Commercialising Economy, England 1086 to c. 1300, pp. 7–27, Manchester: ManchesterUniversity Press.

Stacey, R.C. (1997). Parliamentary negotiation and the expulsion of the Jews from England. Thirteenth CenturyEngland, vol. 6, pp. 77–103.

Stacey, R.C. (2000). ‘Antisemitism and the medieval English state’, in (J.R. Maddicott and D.M. Palliser, eds.),The Medieval State: essays presented to James Campbell, pp. 163–77, London: Hambledon Press.

Stampfer, S. (2003). ‘What actually happened to the Jews of Ukraine in 1648?’, Jewish History, vol. 17(2),pp. 207–27.

Stein, S. (1956). ‘Usury and the medieval English church courts’, Journal of Semitic Studies, vol. 1(2),pp. 141–62.

Stow, K.R. (1981). ‘Papal and royal attitudes toward jewish lending in the thirteenth century’, AJS Review, vol.6, pp. 161–84.

Stow, K.R. (1992). Alienated Minority, the Jews of Medieval Latin Europe, Cambridge, MA: Harvard UniversityPress.

Sutcliffe, A. (2000). ‘Can a Jew be a Philosophe? Isaac de Pinto, Voltaire, and Jewish participation in theEnlightenment’, Jewish Social Studies, vol. 6(3), pp. 31–51.

Tazbir, J. (1973). A State without States, Warsaw: The Kosciuzko Foundaton Twayne Publishers.Tol, R.S. and Wagner, S. (2010). ‘Climate change and violent conflict in Europe over the last millennium’,

Climatic Change, vol. 99(1–2), pp. 65–79.Trachtenberg, J. (1943). The Devil and the Jews, Philadelphia, PA: The Jewish Publication Society.Vidal-Robert, J. (2013). War and inquisition: repression in early modern Spain. CAGE Working Paper

No. 119, University of Warwick.Vital, D. (1999). A People Apart, The Jews in Europe 1789–1939, Oxford: Oxford University Press.Voigtl€ander, N. and Voth, H.-J. (2012). ‘Persecution perpetuated: the medieval origins of Anti-Semitic

violence in Nazi Germany’, Quarterly Journal of Economics, vol. 127(3), pp. 1–54.Waldinger, M. (2014). ‘The economic effects of long-term climate change: evidence from the little ice age,

1500–1700’, memo, London School of Economics.Weinryb, B.D. (1972). The Jews of Poland, Philadelphia, PA: Jewish Publication Society of America.White, S. (2014). ‘The real little ice age’, Journal of Interdisciplinary History, vol. 44(3), pp. 327–52.Yamaguchi, K. (1991). Event History Analysis, Newbury Park, CA: Sage.Yanagizawa-Drott, D. (2015). Propaganda and conflict: evidence from the Rwandan Genocide. Quarterly

Journal of Economics, vol. 129(4), pp. 1633–85.Zagorin, P. (2003). How the Idea of Religious Toleration Came to the West, Princeton, NJ: Princeton University

Press.Zhang, D.D., Jim, C., Lin, G., He, Y.-Q., Wang, J.J. and Lee, H.F. (2006). ‘Climatic change, wars and dynastic

cycles in China over the last millennium’, Climatic Change, vol. 76(3-4), pp. 459–77.



Documents

JEWISH PERSECUTIONS AND WEATHER SHOCKS: 1100 1800*